NERSC is hosting an Advanced NVIDIA HPC SDK Training, led by NVIDIA HPC Solutions Architects Vineeth Gutta and Scot Halverson. The event is scheduled March 4th to 5th, 2026 and NCAR affiliated users are welcome to apply. To note, NERSC’s compute cluster Perlmutter has very similar architecture to Derecho.

NVIDIA will present a two day training session software for High Performance computing developers to build and optimize applications that run on NVIDIA GPU-accelerated systems. We'll cover NVIDIA's HPC SDK, including compilers (C, C++, Fortran), libraries (math, communication), and tools, as well as cover best practices to enable scientists to effectively use NVIDIA software for high performance computing. In addition to the HPC SDK, we'll do a deeper dive on C++ Standard Parallelism, profilers and correctness tools like NSight Systems, NSight Compute, and Compute Sanitizer, as well as a tutorial on NVIDIA's newest GPU Programming model, CuTile.

Registration is now open and the deadline for non-NERSC users is February 20th.